Availability

Availability is usually expressed as a percentage indicating the system’s uptime over a specific period

Formal definition:

Availability = Uptime / (Uptime + Downtime)

Uptime: The period during which a system is functional and accessible

Downtime: The period which a system is unavailable due to failures, maintenance, or other issues

Availability Tiers §

Availability is expressed in “nines”. Higher availability -> less downtime

Each additional “nine” represents an order of magnitude improvement in availability

Strategies for Improving Availability §

Redundancy §

Redundancy involves having backup components that can take over when primary component fails

Techniques:

• Server Redundancy: Deploying multiple servers to handle requests, ensuring that if one server fails other can continue to provide service
• Database Redundancy: Creating a replica database that can take over if the primary database fails
• Geographic Redundancy: Distributing resources across multiple geographic locations to mitigate the impact of regional failures

Load Balancing §

Load Balancing distributes incoming network traffic across multiple servers to ensure no single server becomes a bottleneck, enhancing both performance and availability

Techniques:

• Hardware Load Balancers: Physical Devices that distribute traffic based on pre-configured rules
• Software Load Balancers: Software solutions that manage traffic distribution, such as HAProxy, Nginx, or cloud-based solution like AWS Elastic Load Balancer

Failover Mechanisms §

Failover Mechanisms automatically switch to a redundant system when a failure is detected

Techniques:

• Active-Passive Failover: A primary active component is backed by passive standby component that takes over upon failure
• Active-Active Failover: All components are active and share the load. If one fails, the remaining components to handle the load seamlessly

Data Replication §

Data replication involves copying data from one location to another to ensure data is available even when one location fails

Techniques:

• Synchronous Replication: Data is replicated in real-time to ensure consistency across locations
• Asynchronous Replication: Data is replicated with a delay, which can be more efficient but result in slight data inconsistencies

Monitoring and Alerts §

Continuous health monitoring involves checking the status of system components to detect failures early and trigger alerts for immediate action

Techniques:

• Heartbeat Signals: Regular signals sent between components to check their status
• Health Checks: Automated scripts or tools that perform regular health checks on components
• Alerting System: Tools like PagerDuty or OpsGenie that notify administrators of detected issues

Best Practices for High Availability §

1. Design for Failure: Assume that any component of your system can fail at anytime and design accordingly
2. Implement Health Checks: Regular health checks allow you to detect and respond to issues before they become critical failures
3. Use Multiple Availability Zones: Distribute your system across different data centers to prevent localized features
4. Practice Chaos Engineering: Intentionally introduce failures to test system resilience
5. Implement Circuit Breakers: Prevent cascading failures by quickly cutting off problematic services
6. Use Caching Wisely: Caching can improve availability by reducing load on backend systems
7. Plan for Capacity: Ensure your system can handle both expected and unexpected load increases